The New Method / Substep / Concept Idea Thread

[Pyjam]

Is there an appropriate thread for this to go into, because I can’t find one..,

Yes, 5th "Sticky threads", just above.
Please, post an example.

 

[GAN 356 X]

Sub-8 with this new method now!

 

[PetrusQuber]

I’m waiting for somebody to say:
Or you could just use Petrus

 

[PetrusQuber]

R U’ L2 F2 B’ D2 R D2 B2 L’ D F’ B U2 R’
y’ x’// inspection
L2 R U L U’ R B2 D’// bad 2x2x2 8 HTM, 8 HTM
R’ F2 R2 F2 R2 U F’// Psuedo block, wrong 2x2x1 7 HTM, 15 HTM
U’ R U2 R’ U2// Roux block complete 5 HTM, 20 HTM
B’ R F’ B// Edge insertion 4 HTM, 24 HTM
R’ U R U2 R’ U R// F2L pair 7 HTM, 31 HTM
y2 R U2 R2 U’ R2 U’ R2 U2 R// Chameleon Headlight 9 HTM, 40 HTM
y F2 U L R’ F2 L’ R U F2// U Perm 9 HTM, 49 HTM
U’// AUF 1 HTM, 50 HTM

Of course, there was a edge OLL skip, and an accidental COLl, but I believe this method could be optimised further with better block building at the start, D layer EO influencing and forced edge OLL skips.

 

[Cubinwitdapizza]

R U’ L2 F2 B’ D2 R D2 B2 L’ D F’ B U2 R’
y’ x’// inspection
L2 R U L U’ R B2 D’// bad 2x2x2 8 HTM, 8 HTM
R’ F2 R2 F2 R2 U F’// Psuedo block, wrong 2x2x1 7 HTM, 15 HTM
U’ R U2 R’ U2// Roux block complete 5 HTM, 20 HTM
B’ R F’ B// Edge insertion 4 HTM, 24 HTM
R’ U R U2 R’ U R// F2L pair 7 HTM, 31 HTM
y2 R U2 R2 U’ R2 U’ R2 U2 R// Chameleon Headlight 9 HTM, 40 HTM
y F2 U L R’ F2 L’ R U F2// U Perm 9 HTM, 49 HTM
U’// AUF 1 HTM, 50 HTM

Of course, there was a edge OLL skip, and an accidental COLl, but I believe this method could be optimised further with better block building at the start, D layer EO influencing and forced edge OLL skips.

I believe i understand this. Also, really cool method! I think what you could also do is instead of inserting edges and then do F2L pair you could just insert a pseudo pair or just insert the edge and then do tols.

 

[PetrusQuber]

I believe i understand this. Also, really cool method! I think what you could also do is instead of inserting edges and then do F2L pair you could just insert a pseudo pair or just insert the edge and then do tols.

Yeah, the edge thing takes up too many moves...

 

[PapaSmurf]

Edge+EO shouldn't take too many moves and could be worth pursuing.

 

[KAINOS]

One of the problems that 2-gen redux methods have is that the 2-gen part of the solve is actually not that efficient. Traditional method (ZZ/Petrus F2L+2GLL) gives you something like 28~30 moves(13~15 for F2L, 14.65 for 2GLL (13.15 for algs+AUF)), when the average optimal is around 16~17. So I'm trying to find better alternate method that could finish the solve in <25 moves consistently.

One of the better methods I've come up with is this: Solving tripod(two 2x2x1 blocks - one on DBR and the other on U) and then solving the rest with 1 algorithm. (~300 algs in total) The latter part seems to be slightly better or at least equal to normal 2GLL in both movecount and ergonomics, and some of the algs are really short and nice (~9-12 moves) because of the open FR slot. Also making tripod does solves 1 more piece than F2L, but you have 4 different options for block on U and blockbuilding could be easier as well because of, again, the open slot.

That being said, it would only save only 2 moves at absolute best which is not sufficient to be under 25 moves, of course. And any other stuff I've found couldn't really top this, either. That's why I posted here - I want to hear some new possible ideas from other people. What do you guys think about it? Are there any good 2-gen solving method you can think of? Or do you think it would be impossible to meet a such goal?

 

[Etotheipi]

One of the problems that 2-gen redux methods have is that the 2-gen part of the solve is actually not that efficient. Traditional method (ZZ/Petrus F2L+2GLL) gives you something like 28~30 moves(13~15 for F2L, 14.65 for 2GLL (13.15 for algs+AUF)), when the average optimal is around 16~17. So I'm trying to find better alternate method that could finish the solve in <25 moves consistently.

One of the better methods I've come up with is this: Solving tripod(two 2x2x1 blocks - one on DBR and the other on U) and then solving the rest with 1 algorithm. (~300 algs in total) The latter part seems to be slightly better or at least equal to normal 2GLL in both movecount and ergonomics, and some of the algs are really short and nice (~9-12 moves) because of the open FR slot. Also making tripod does solves 1 more piece than F2L, but you have 4 different options for block on U and blockbuilding could be easier as well because of, again, the open slot.

That being said, it would only save only 2 moves at absolute best which is not sufficient to be under 25 moves, of course. And any other stuff I've found couldn't really top this, either. That's why I posted here - I want to hear some new possible ideas from other people. What do you guys think about it? Are there any good 2-gen solving method you can think of? Or do you think it would be impossible to meet a such goal?

would it be at all beneficial to solve eo before the last tripod 2x2x1? It would make the block a bit more efficient and reduce the alg count, but I don't know how much good it would do.

 

[PapaSmurf]

You need EO solved for it to be 2gen, so he's only talking about EO solved.

 

[ProStar]

A (probably wrong) thought about improving CFOP F2L
This is most likely wrong, but would using Roux's F2B(First 2 Blocks) to start and then adding the extra 2 edge pieces by utilizing the M slice to finish F2L, then finishing the cube with OLL and PLL be faster? Or would that be slower than doing F2L normally?

 

[AbsoRuud]

CFOP movecount:
Cross, F2L, OLL, PLL

Roux movecount:
FB, SB, CMLL, LSE

Your method: FB, SB, something something, OLL, PLL.

It seems to me that the 2-3 moves that you'd be doing more when doing cross right away would be less than 'something something'.

 

[ProStar]

CFOP movecount:
Cross, F2L, OLL, PLL

Roux movecount:
FB, SB, CMLL, LSE

Your method: FB, SB, something something, OLL, PLL.

It seems to me that the 2-3 moves that you'd be doing more when doing cross right away would be less than 'something something'.

But moves isn't the only thing that counts, some moves are less efficient than others. I've been trying to find splits for F2B compared to Cross + F2L for an average solver, but I can't find them.

 

[Matt11111]

Disclaimer: I'm not color neutral, nor am I particularly good at lookahead, so I'm probably the least qualified person to figure out whether this idea is any good or not.

So I was screwing around on 3x3, and a really weird idea came to mind. After doing your cross (let's say you do white) and two adjacent F2L pairs (so for example the two green pairs), you have a 2x2x3 block plus your last white cross edge. What if, in addition to looking for your next pair, you tried to locate the yellow-green edge, and if you can make a green cross and the next green pair easily, you stop solving on white and rotate to do green instead?

Not much progress would be lost since the only piece that might get moved out of position is the last white cross edge, but I wonder if looking for F2L pairs on two different colors might be too much to be thinking about when you're trying to get a fast solve.

So what do you guys think? Could this idea have any potential? Like I said, I'm not color neutral, so it's definitely not an idea that could benefit me, but maybe some color neutral folks could provide some input.

 

[Aerma]

It's an interesting idea, but I would think that it would take a lot of effort to get decent at, with very little reward... might just be easier and help look-ahead to just stick with the same F2L. I'm not color neutral though, so maybe someone else thinks it's a better idea!

 

[PapaSmurf]

Everyone has made that suggestion before a=nd everyone realises that it's worse than CFOP and it's worse than Roux. F2L is pretty fast anyway, and you're removing the benefits of CMLL+LSE for a worse system.

 

[Cubingcubecuber]

Everyone has made that suggestion before a=nd everyone realises that it's worse than CFOP and it's worse than Roux. F2L is pretty fast anyway, and you're removing the benefits of CMLL+LSE for a worse system.

Has anyone ever suggested 3/4 cross with missing one in front, F2L(made rotationless and more efficient because the M slice is almost free), CxLL?(Doesn’t need to preserve EO, but the M slice isn’t completely free), L5E(EO is only 5 cases because on solved edge in DB, the rest can be done using one of 16 algs{4 are EPLL, and all except the 5 cycles can be EPLL with setup moves})(or you could do L5EOP and EPLL)(or FD and ELL)




Edit: this is basically Russo

 

[Filipe Teixeira]

lars petrus had this idea when developing petrus.

You have three corners to choose from. It's important to pick the easiest, just like in step 1.

Source:https://lar5.com/cube/fas2.html

I think it's doable if you can recognize cases quickly, but it would be some wasted moves if the cross is done.
If you go directly for 2x2x3, call it freefop and you're good to go

 

[PapaSmurf]

Yes, same story. Do Roux.

 

[Etotheipi]

Would it be at all beneficial for those who don't know full CMLL to, in Roux, after FB, do SB 2x2x1, CP, finish SB, then 2-gen CMLL? Recog is pretty bad for CP, but you can (I think) always solve it with at most 2 sledgehammers and some AUFs, and then you only need one alg out of a set of seven to one look CMLL. Obviously full CMLL would be better, but for those who don't know it would this help?